基于随机森林方法的遥感影像耕地提取与变化分析

Extraction and Change Analysis of Cultivated Land from Remote Sensing Images Based on Random Forest Method

耕地是人类社会稳定发展的物质基础,是粮食生产的重要前提与保障。耕地的变化检测对于耕地的保护和利用都具有十分重要的意义。该文以南京市江宁区为研究区,使用2017年、2019年、2021年3期哨兵2号影像,采用随机森林分类方法,提取耕地信息并分析其时空变化。结果表明:随机森林分类方法能有效识别耕地,分类精度均在80%以上。2017—2019年间,江宁区耕地面积变化不大;2021年与2019年相比,耕地面积略有减少,部分耕地转化为其他用地类型。在空间上,西部、东南部地区的耕地与其他用地类型间转化面积较大;总体上,耕地重心不断向东南方向迁移。耕地的时空变化受多种因素的影响,经济发展与城市扩张是其中的主要因素,及时准确地获取耕地变化信息可为相关管理部门提供科学依据。

Cultivated land is the material basis of stable development of human society,and is an important premise and guarantee of food production.The change detection of cultivated land is very important for the protection and utilization of cultivated land.In this paper,jiangning District,Nanjing was selected as the research area,and the sentry 2 images of 2017,2019 and 2021 were used to extract cultivated land information and analyze its spatial and temporal changes by random forest classification method.The results show that the random forest classification method can effectively identify cultivated land,and the classification accuracy is above 80%.From 2017 to 2019,the cultivated land area in Jiangning District did not change much.In 2021,compared with 2019,the area of cultivated land decreased slightly,and part of cultivated land was converted to other land types.In terms of space,the conversion area between cultivated land and other land types is large in the west and southeast regions.On the whole,the center of gravity of cultivated land moves southeast continuously.The spatial-temporal change of cultivated land is affected by many factors,among which economic development and urban expansion are the main factors.Timely and accurate acquisition of cultivated land change information can provide scientific basis for relevant management departments.